Background: Although reverse sequence algorithms (RSA) for syphilis screening are performing well, they still have to rely on treponemal confirmatory tests at least for sera reactive by enzyme immunoassay/chemiluminescence immunoassay (EIA/CIA) and unreactive by rapid plasma reagin (RPR). Quebec's laboratory network previously showed that 3.3% of EIA/CIA reactive and weakly-reactive RPR samples (RPR titer of 1 to 4) would have been misclassified as syphilis cases if a treponemal confirmatory test had not been performed.
Objectives: To correlate the magnitude of signal-to-cutoff (S/CO) ratios of the 4 most used commercial first-line EIA/CIA kits in Quebec with syphilis confirmation results and establish a S/CO value above which treponemal confirmation would not be required.
Methods: Serum samples from previously undiagnosed individuals (n = 7 404) obtained between January 2014 and February 2017 that were reactive by EIA/CIA and either negative by RPR or reactive with a low titer (1 to 4) were included in the study. All samples were tested with Treponema pallidum particle agglutination (TP-PA) and, if negative or inconclusive, with a line immunoassay (LIA). Syphilis infection confirmation was defined by a reactive TP-PA or LIA. Logistic regression analysis was used to determine S/CO values (95% CI lower bound = 0.98) above which confirmation would not be required. The four kits studied were Architect TP, BioPlex IgG, Syphilis EIA II, and Trep-Sure.
Results: Of 2609 reactive EIA/CIA specimens tested for the determination of S/CO values, 1730 (66%) were confirmed as true syphilis cases. Confirmation rate was significantly higher in samples with low-titer positive RPR (92%) than with negative RPR samples (54%); p<0.01. A linear probability model (95% CI lower bound = 0.98) predicted the S/CO value above which a confirmation would no longer be needed for the Architect TP (16.4), Bioplex IgG (7.4) and Trep-Sure (24.6). No linearity was observed between the S/CO value of Syphilis EIA II and the confirmation rate. The validity of the predicted S/CO values was investigated using 4 795 specimens. The use of an S/CO value of 16.4 with the Architect TP kit and of 24.6 for the Trep-Sure kit would obviate the need for confirmation of 18.5% and 13.2% of sera from the all RPR subgroup, respectively. For the BioPlex IgG kit, 81.1% of sera would not require confirmation when using the S/CO value of 7.4 in the low titer RPR subgroup.
Conclusion: Signal-to-cut-off values could be used to identify sera that do not require extra treponemal confirmation for 3 of the 4 most used first-line EIA/CIA kits in Quebec. Using these values in our current reverse screening algorithm (RSA) would avoid the need for confirmatory tests in 14 to 20% of sera, a proportion that could reach 75% among low-titer RPR.